FAQ: Data ocean slab mode (DOCN-SOM)

[hanj@pku_edu_cn]

I had read CESM1.0 user guide and those files in /pop_som_frc folder before asking questions. I certainly know what T31_g37 is. I do not know why I cannot use a same resolution of SOM forcing file to perform simulations. Should SOM forcing files be interpolated to a regular grid for any resolution simulations?These things cannot be found in a user guide. Cheers,Jing

 

[dbailey]

I just set up a sample test using the CESM1.2.2 code base as follows:./create_newcase -compset E1850CN -mach yellowstone -res T31_g37 -case ~/cases/somtestcd ~/cases/somtest./cesm_setup./xmlchange DOCN_SOM_FILENAME=pop_frc.gx3v7.110128.nc./somtest.build./somtest.submitIt runs 6 days with no trouble.

 

[hanj@pku_edu_cn]

All right. It work for CESM1.2.2.However, I'm using CESM1.0.5. I set up a test like that:./create_newcase -compset E1850CN -mach yellowstone -res T31_g37 -case */som_trycd */som_try./configure -casecd Buildconfchanging docn.buildnml.csh as:           $DIN_LOC_ROOT = /glade/p/cesm/cseg/inputdata/           set mod_dompath = $DIN_LOC_ROOT/ocn/docn7           set mod_domfile = domain.ocn.gx3v7.090903.nc           set dat_dompath = $DIN_LOC_ROOT/ocn/docn7/SOM           set dat_domfile = pop_frc.gx3v7.110128.nc           set dat_datpath = $DIN_LOC_ROOT/ocn/docn7/SOM           set dat_datfile = pop_frc.gx3v7.110128.nc./som_try.yellowstone.build./som_try.yellowstone.submitUnfortunately, the model gave the error "Xc should be regular and increase monotonically" in ocn.log. Cheers,Jing  

 

[dbailey]

Ok. This is much clearer now. I would recommend moving to CESM1.2.2, however if you absolutely must use CESM1.0.5 then you need to use a different domain file:set mod_dompath = $DIN_LOC_ROOT/ocn/docn7/SOMset mod_domfile = pop_frc.gx3v7.110128.nci.e. use the SOM forcing file for your domain information. I had to dig bag several years, but the gx3v7 domain definition changed ever so slightly. So, the DOCN thought it had to remap from gx3v7 to gx3v7 which it shouldn't.

 

[hanj@pku_edu_cn]

Great.I prefer to the relative old version because I have a long-term B1850 simulation using CESM1.0.I have tried this method. It works.I really appreciate your help. Best regards,Jing

 

[pdn@hawaii_edu]

Hi Dave, I'm running an E case using CESM1.2.1 with CAM4 physics and f45_gx37 grid. I computed the Q-flux from a B control ran with the exact same setup (CAM4 physics, same grid). My problem is: the climatology of the slab control does not agree that of the coupled run, mainly over the high latitudes due to reduced sea ice extent. I would welcome any advise on how to diagnose the cause of the discrepancy between the two climates. I expected them to be the same. Thanks!

 

[dbailey]

Hi Pedro,I have no experience with running the SOM at the f45_g37 resolution combination. I have run with the T31 CAM atmosphere, but not the f45. The T31 CAM atmosphere produces very extensive and unrealistic sea ice. How does it look with the f45 atmosphere? The SOM configuration does generally have less extensive ice and a warm bias in the SST of 0.1-0.3C. What are the default shortwave parameters in ice_in? I'm looking for R_snw, dT_mlt, and rsnw_mlt in particular. Are these the same for your fully coupled run as well as the SOM run? Was the fully-coupled run an 1850 control or a present-day control?

Dave

 

[dbailey]

Also, how many years are you averaging over for the SOM forcing?Dave

 

[pdn@hawaii_edu]

I didn't know about the excessive ice extent in the T31 CAM runs. My simulations are idealized, so I'm not overly concerned about the realism of the sea ice extent. Howver, I'm concerned about the different sea ice extent of the slab run relative to the coupled control. I get differences in ice coverage (ICEFRAC) of 20% and SSTs of 5 K in the slab relative the coupled case. The slab case has less sea ice and warmer high latitudes (I can send you a plot if you want to look at the spatial extent, the reductions are at the sea ice edge). I compared the ice_in namelists and they are the exact same, the paramenters you mention are:r_snw =  -2.00dt_mlt_in =  2.00rsnw_melt_in =  2000.The coupled run was a B1850 compset with CAM4 physics. For the slab run I started from an E1850C5 case and modified the physics to be CAM4. I did this by changing the following parameters in env_run.xml:CAM_CONFIG_OPTS"   value="-phys cam4"CLM_NAMELIST_OPTS"   value=""Maybe I'm missing some other settings to go from CAM5 physics to CAM4... would that affect CICE?To answer your last question, I computed the Q-flux from the last 100 years of a nearly equilibrated control.

 

[dbailey]

Oh wait, did you select the E compset (default E2000) or the E1850 compset? Check your CO2 value and I bet that is it. An SST difference of 5K could only be CO2 methinks.Dave

 

[pdn@hawaii_edu]

I used the E_1850_CAM5 compset so the CO2 are correct. I actually used the same namelists for CAM,CLM, and RTM than my coupled run. Looks like the Q-flux is not doing the right thing under the ice. I computed the Q-flux using the pop_frc_mlt.ncl. The version of that script that I'm using computes the surface heat flux as SHF + QFLUX, however other variables are loaded, for instance MELTH_F. I don't fully understand what the difference between QFLUX and MELTH_F so I thought I double check with you. 

 

[dbailey]

Ok. Let's take this offline (dbailey@ucar.edu) and you can point me to your code on yellowstone. I find it hard to believe that the SOM forcing could generate a 5K difference.Dave

 

[1475922792@qq_com]

Hello, have you solved the problem? I have encountered the same problem like you. I  set the same compset and use the same SOM data as you did. How did you solve the problem at that time?zhao

 

[aamir003@ucr_edu]

Hi all,I ran the SOM by default for 10 years and did not face any problem after that I've tried to run SOM simulation with my observational data (my data has f19_f19 resolution 144x96). first I created a case: create_newcase -case E_1850_CN -res f19_g16 -compset E_1850_CN -mach yellowstone and then added this lines to the env_run_xml: After that set up and build the model, but when I submit the job after a couple of minutes job exited and I got this error:   1: Opened file E_1850_CN_aerosol.cam.h1.0001-01-01-00000.nc to write      458752   1: pio_support::pio_die:: myrank=          -1 : ERROR:   1: pionfwrite_mod::write_nfdarray_double:         234 :   1: NetCDF: Numeric conversion not representable   1:Abort(1) on node 1 (rank 1 in comm 1140850688): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 1   1:INFO: 0031-306  pm_atexit: pm_exit_value is 1.INFO: 0031-251  task 1 exited: rc=1ERROR: 0031-300  Forcing all remote tasks to exit due to exit code 1 in task 1  88:forrtl: error (78): process killed (SIGTERM)  88:Image              PC                Routine            Line        Source  88:libpthread.so.0    00002AD2F74EA585  Unknown               Unknown  Unknown  88:libpoe.so          00002AD2FC4D4AE2  Unknown               Unknown  Unknown  88:libpthread.so.0    00002AD2F74E2AA1  Unknown               Unknown  Unknown  88:libc.so.6          00002AD2F9A9693D  Unknown               Unknown  Unknown   88:INFO: 0031-306  pm_atexit: pm_exit_value is 1.Thanks in advance for your help. Ana

 

[11438023@zju_edu_cn]

Hi,Can I ask about how cice and SOM are coupled in the code level?The script used to update sst, sss, etc. is found in docn_comp_mod.F90, I beleive this is how open ocean works. However, for grid points that are partly or fully occupied by sea ice (sea ice fraction

 

[dbailey]

SST under the ice is -1.8C by definition. The open ocean temperature is computed by the DOCN SOM mode as you suggest. The atmosphere never sees the temperature under the ice. It only sees the ice surface temperature. Dave

 

[11438023@zju_edu_cn]

Thanks for your response! I still have some confusions about the calculation processes. For my understanding:(1) For open ocean where sea ice fraction is 0, SOM receives energy fluxes (i.e., SW, LW, SH, LH…) from atmosphere and compute SST change and the freezing/melting potential for CICE;(2) For grid points where sea ice fraction is 1, SOM receives penetrating shortwave flux and net heat flux (fhocn) from sea ice. As you said, SST is defined as -1.8deg but still computes the freezing/melting potential;(3) For grid points where sea ice fraction is between 0 and 1, both (1) and (2) happen.  I found DOCN_COMP_MOD.F90 is very simple and only includes the calculation process for (1). I am confused where (2) and (3) are done (in CICE, SOM, or coupler)?  Another question is, the defualt time step is 30mins for CAM and CLM, and 1 hr for CICE. What is the time step for SOM?  PS, There is a script named ice_ocean.F90 in CICE which is reasonable for above cases, but the guider says it’s for stand-alone mode. Thanks!

 

[dbailey]

Think of the DOCN as a "replacement" for the POP active ocean model. Where there is open ocean, the SST and freeze/melt potential are computed by the ocean component. The fluxes over the open ocean are computed by the driver/coupler. These are merged based on ice fraction to send to the atmosphere. So, if a grid cell is 100% ice covered, the atmosphere only sees the fluxes over the ice. The fluxes at the base of the ice come from the ocean component, etc. You are correct the ice_ocean.F90 is not used in the coupled model.Dave

 

[11438023@zju_edu_cn]

I understand how atmopshere works with ice and open ocean, I just don't know how ocean and ice coordinate with each other, especially the energy balance calculation and the freezing/melting processes at the base of the sea ice... You said that "The fluxes at the base of the ice come from the ocean component, etc.", where can I find this calcualtion process or which code includes it?      That is, for grid cells that are 100% ice covered, ice stil passes energy to ocean (shortwave and fhocn), but I cannot find any clus that ocean takes use of these terms...Best,   

 

[dbailey]

Again, this is similar to what POP receives. The coupler merges fluxes from the ice and atmosphere based on the ice fraction. The DOCN receives net shortwave which is:swnet = frac*swpen + (1-frac)*swnetwhere swpen is the penetrating shortwave through the ice and the net absorbed at the open ocean surface. No longwave gets through the ice, but there is a sensible heat exchange from the bottom of the ice. The one difference is that the SOM does not compute salinity/fresh water, so these fluxes from the ice are ignored.Dave